topic Re: PROC GLM vs. PROC REG with categorical predictor in Statistical Procedures

PROC GLM vs. PROC REG with categorical predictor

tam_el — Sat, 24 Feb 2018 15:06:56 GMT

I confused between PROC GLM and REG, I read about the difference but still don't have the ability to answer a question like:

Q: A linear model has the following characteristics:
- A dependent variable (y)
- Three continuous predictor variables (x1-x3)
- One categorical predictor variable (c1with 3 levels)

Which SAS program fits this model?

My answer is C, but I the right answer is D. Why?

Re: PROC GLM vs. PROC REG with categorical predictor

PeterClemmensen — Sat, 24 Feb 2018 15:23:07 GMT

Because, as you can see from the Syntax page of the PROC REG Documentation, PROC REG does not support a class statement.

Re: PROC GLM vs. PROC REG with categorical predictor

tam_el — Sat, 24 Feb 2018 15:32:55 GMT

Thank you, Understood now, I should explicitly create a dummy variable for categorical predictor before applying PROC REG.

Re: PROC GLM vs. PROC REG with categorical predictor

PaigeMiller — Sat, 24 Feb 2018 16:14:09 GMT

@tam_el wrote:
Thank you, Understood now, I should explicitly create a dummy variable for categorical predictor before applying PROC REG.

No.

You should use PROC GLM. Much simpler, you do not have to create dummy variables when you use PROC GLM, and you get the exact same answer as if you created the dummy variable and then ran PROC REG.